High prevalence of epilepsy in Northern Rwanda: Exploring gender differences

Abstract Introduction In sub‐Saharan Africa (SSA), the prevalence of lifetime epilepsy varies widely between subregions and is higher in rural compared to urban regions. Observed versus expected numbers of patients with epilepsy (PwE) in the northern province of Rwanda did not match the prevalence of 49‰ reported in 2005 in Rwanda. We report a confirmatory prevalence study focused on gender‐specific observations. Methods A cross‐sectional door‐to‐door approach was used in three rural villages. First, epilepsy screening using the Kinyarwanda version of the Limoges questionnaire was performed. Second, confirmation of epilepsy diagnosis was completed by trained physicians. Results In total, 2681 persons (56.14% female) were screened. Of 168 positively screened, 128 persons were diagnosed with epilepsy confirming the prevalence of lifetime epilepsy of 47.7‰ (CI 39.8–56.8). The diagnosis gap was 62.5% with 80 newly diagnosed. The overall female:male ratio was 1.61:1.00. A male preponderance below 9 years of age inverted to a female preponderance above 20 years of age. Female PwE had an older age at first seizure, reported different reasons for not seeking care, and differed from male PwE in possible etiology. For previously diagnosed PwE, the treatment gap was more than 77%. Conclusion A high prevalence in rural areas was confirmed, with an observed female/male ratio among the highest of published door‐to‐door surveys in SSA. Gender differences in associated co‐morbidities and age at first seizure warrant future research of underlying etiologies and possible survival bias. A better understanding and focus on gender‐associated care‐seeking patterns, education, and specific needs are recommended.


INTRODUCTION
Epilepsy is characterized by unpredictable, recurrent seizures and may have cognitive, psychological, and socio-economic consequences . This common chronic disease is affecting over 70 million patients living with epilepsy (PwE) worldwide, of whom nearly 85% live in low-income and low-and middle-income countries (LLMIC) (Ngugi et al., 2010). It contributes nearly to 1% in the global burden of disease and 20% of the global burden of epilepsy measured in disability-adjusted life years (DALYs) in sub-Saharan Africa (SSA) (GBD 2015Neurological Disorders Collaborator Group, 2017. The estimated prevalence of epilepsy varies widely between highincome counties (range 4 to 7‰) and LLMIC (range 5 to 74‰) (Ngugi et al., 2010;Preux & Druet-Cabanac, 2005). A recent meta-analysis of 38 studies in 19 SSA countries estimated an overall prevalence of active and lifetime epilepsy at 9‰ and 16‰ respectively, varying widely between subregions but with a consistent trend of higher epilepsy prevalence in rural compared to urban settings (Owolabi et al., 2020). The epilepsy prevalence in Rwanda was estimated at 49.3‰ in a community-based door-to-door survey in 2005, also indicating higher prevalence in rural compared to urban areas (Sebera et al., 2015).
Rwanda is a landlocked country in East Africa and home to more than 12.5 million inhabitants, with a mean age of 20 y, of which less than 20% live in urban areas (Nyirandagijimana et al., 2017). A communitybased health insurance model provides cost coverage. Mental health care in Rwanda, including epilepsy care, has a leveled pyramidal structure, ensuring decentralization in all health facilities, ranging from primary level health centers under supervision of secondary district or province hospitals to tertiary reference hospitals (Nyirandagijimana et al., 2017). On average in rural areas, health centers provide care to the population of 5.2 sectors, serving in turn 6.9 villages per sector (Rwanda Administration, n.d.). With three neurologists in Rwanda access to neurology care is limited.
During a feasibility assessment for a study on epilepsy and comorbid depression in an urban and a rural setting, the observed number of PwE at health centers in rural areas of the northern province was unexpectedly low compared to our estimates based on a previously reported prevalence of 49‰, triggering a confirmatory study of epilepsy prevalence in rural areas in Rwanda (Sebera et al., 2015).
We report the results of a door-to-door survey estimating epilepsy prevalence in three villages of the Musanze Province in Northern Rwanda, using the Limoges epilepsy screening questionnaire (Diagana et al., 2006;Preux et al., 2003).

Study area, conduct, and sample size
We conducted a cross-sectional study with a door-to-door approach in three rural villages, that is, Kaberege (supervised by Gataraga Health Center), Mwidagaduro, and Rutemba (supervised Karwasa health center) in the Musanze District (Northern Province, Rwanda). The distance from Kaberege, Mwidagaduro, and Rutemba villages to the Ruhengeri Reference Hospital is 13, 6, and 4 km, respectively ( Figure 1). These villages were selected by the directors of the Ruhengeri referral hospital and health centers in agreement with the village elders based on total population, distance to the health centers, rural character, and absence of previous epilepsy awareness campaigns The low number of PwE observed in the regions called for an assumption on the possible prevalence in order to calculate the required sample size. We used a conservative prevalence of 10‰, in line with another large study in sub-Saharan Africa, which was only 20% of the previously reported Rwandan prevalence. A sample size of 2000 persons was calculated with a precision of 0.5% and a 97.5% confidence interval (CI) (Kohn & Senyak, 2021;Sebera et al., 2015).
We used the 2017 census data as reference for the total population, provided by the CHW supervisors during preparatory meetings. The census data were not broken down by gender or age groups.

Screening questionnaire and data collection
The Limoges epilepsy screening questionnaire "part 2" was used for screening for epilepsy, which had previously been translated into Kinyarwanda and adapted to the sociocultural context of Rwanda (P.-M. Preux et al., 2003;Sebera et al., 2015). The questionnaire was initially validated in Mauritania with sensitivity and specificity of 95.1% and 65.6% respectively (Diagana et al., 2006).
We used electronic data capturing, avoiding double data entry and possible paper data loss. An application for mobile devices, Android version 4.0 and above, was built reflecting the questionnaire. Data from the screening questionnaire were collected on a Rwanda-based server, maintained by One Family Health, a non-governmental organization based in Kigali, Rwanda.
Data on demographics, clinical characteristics, treatment, and technical investigations of positive screened subjects, were collected in an Excel spreadsheet.

Survey process
The door-to-door survey process, involved three phases.

Data analysis
Descriptive analysis was performed on anonymized data. Seizure classification according to the 2017 seizure classification was performed by both the senior resident and a general neurologist after review of the available documentation (Fisher, Cross, D'Souza, et al., 2017;. In case of discrepancies between physicians, a consensus meeting provided a final outcome. Data were analyzed using Excel functions; t-tests for comparison of means and Chi-squared test were used for intergroup differences. Prevalence was calculated using sample-size.net (Kohn & Senyak, 2021).

Ethical review committee and local health and administrative authorities
The study was conducted within the framework of the investigational epilepsy and co-morbid depression study approved by the College

Literature review
To evaluate the consistency of the observed gender differences, a literature search in Medline, PubMed, French and Portuguese literature on "prevalence," "epilepsy," "sub-Saharan Africa," and "screening" was performed. Female/male ratios of PwE were tabulated only in case of full manuscripts with a door-to-door design and if prevalence by gender was reported.

Prevalence and diagnosis gap
The study was conducted in September and October 2017.

Demographics of PwE
Striking differences in the demographics of PwE when analyzed by gender were observed ( Table 2). The overall female:male ratio was 1.61:1.00. There was a male preponderance of PwE aged 9-year and below inverting to a female preponderance in the age group 20-years and above, also reflected in a large intergender difference in mean age.
A new diagnosis of epilepsy was made in nearly 7 in 10 of all female PwE, compared to only 5 in 10 male patients (

Epilepsy characteristics
Age at onset of first seizure differed markedly between male and female PwE. Comparing newly diagnosed to previously diagnosed PwE, a trend of epilepsy onset at an older age for newly diagnosed female PwE was observed (Tables S1 and S2).

Treatment
Of 48  Reasons for discontinuation of any AED treatment were cited by 33, of whom 16 female patients: seizure cessation in 11/33 (33.3%), lack of efficacy in 10 (30.3%) and beliefs such as "epilepsy is untreatable" in 3 (9.1%) PwE. Of 11 patients citing seizure cessation as reason for discontinuation, only two were actually seizure-free for more than a year upon clinical exam whilst others had ongoing seizures. Gender differences for reason of discontinuation between men and women included beliefs and ignorance more frequently cited by female PwE compared to low seizure impact and perceived AED inefficacy by males (Figure 2).

Gender differences-Literature review
Our literature search resulted in 39 manuscripts of which eight had a door-to-door design and reported gender-specific prevalence of lifetime and active epilepsy.
One manuscript reported prevalence in five distinct sub Saharan regions, represented separately in Table 4. Male/female prevalence in different studies was highly variable, ranging from 3.2 to 123‰ for males and 2.5 to 90‰ for females. Also, gender ratio varied markedly.
Of note are the differences in study design, case definitions, and study populations, for example, age cut-off, active versus lifetime epilepsy, and adjusted versus crude prevalence rate. All studies, with an exception of the Benin study, were conducted in rural areas.  (Sebera et al., 2015). Prevalence varied across villages, yet this was not statistically significant.

Prevalence and gender
This prevalence of lifetime epilepsy is high compared to a recently published meta-analysis on prevalence of (lifetime and active) epilepsy in SSA, with an overall 9.6‰ and 4.8‰ prevalence in rural and urban settlements, respectively (Owolabi et al., 2020). Our prevalence is even higher than the highest prevalence of 30.2‰ in Central African countries (Owolabi et al., 2020). It is also higher than previously reported in Rwanda using a one epilepsy question screening tool in a cross-sectional study (Simms et al., 2008). It is considered that cross-sectional studies may underestimate the prevalence of lifetime epilepsy by 75%, and of non-convulsive epilepsy by up to 50% (Ngugi et al., 2013). We deem it unlikely that our data largely underestimate prevalence in this region given the good sensitivity of the Limoges questionnaire as previously demonstrated and as we identified 37 in 128 (28.9%) PwE with non-motor seizures, which are difficult to identify. In addition, the positive predictive value of our screening was similar to other door-to-door surveys (Dent et al., 2005;Ndoye et al., 2005;Njamnshi et al., 2008).
In our cohort, a significant gender difference with high number of total and newly diagnosed female PwE was observed. In SSA, reported gender-specific prevalence of epilepsy is highly variable (Table 4), with higher male or with higher female prevalence across age groups and geographies. Our results illustrate a female:male ratio which inverted with age, exceptionally low at 0.35:1.00 in the 9 years or below age group and high at 3.60:1.00 in the age group above 40 years-of-age.
This age-specific prevalence inversion has been observed in other SSA countries with a male predominance in the age group < 15 years reported in several door-to-door and hospital-based surveys (Atugonza et al., 2016;Bistervels et al., 2016;Lagunju et al., 2016;Matonda-Ma-Nzuzi et al., 2019). One study reported a female:male ratio as low as 0.64:1.00 . The female:male ratio of 3.60:1.00 in PwE aged 40 years and above is markedly higher when compared to other surveys with a high female:male ratio of up to 2.00:1.00 (Birbeck & Kalichi, 2004;Coleman et al., 2002;Ezeala-Adikaibe et al., 2016;Mwangala et al., 2018). Our observations prove consistent with age of onset of first seizure when comparing all and newly diagnosed PwE, with possible explanations, including age-specific aetiologies, care seeking patterns, degree of self-care, gender-dependent risk factors, competing mortality risks, or even a study bias due to case ascertainment (Birbeck & Kalichi, 2004;Winkler et al., 2009).

Epilepsy characteristics and gender specifics
Possible gender-specific factors relative to aetiology and epilepsy characteristics were further explored.
Seizure-onset classification according to the ILAE 2017 guidelines nor seizure frequency were different between sexes. Age of first seizure differed in men versus women, probably only reflecting observed differences in age group distribution.
Head injuries, up to 15.4% of all reported comorbidities, were reported by 14.3% (7/49) of male PwE. This is in line with 15.2% of PwE more than 18 y reporting head injuries in Nigeria, yet high considering 60% of our male population was less than 20 y (Lagunju et al., 2016). In this age group, frequencies of head injury varied from 3.6%, 3.9-4% to 8-8.8% (Burton et al., 2012;Kakooza-Mwesige et al., 2017;Ngugi et al., 2013), suggesting a trend towards increased number of head injuries with age.
The factor of having a first-degree relative living with epilepsy did not differ by gender at 14% and is in line with 11.6%, 13%, and 14.5% reported in Tanzania, South Africa, and Uganda, respectively (Ackermann et al., 2019;Burton et al., 2012;Kakooza-Mwesige et al., 2017).
Co-morbidity or history of infectious diseases affecting more than 10%, was similar for both genders and in line or slightly lower (range 5.5%-13%) than in other SSA regions (Assadeck et al., 2019;Bistervels et al., 2016;Burton et al., 2012;Carter et al., 2004;Kariuki et al., 2014;Samia et al., 2019). There is a clear association between epilepsy and HIV with newonset seizures occurring in up to 11% of HIV infected persons (Howlett, 2019). The observed 4.6% of a HIV positive status was somewhat higher than the percentage of 1.7% observed in a Rwandan tertiary epilepsy center, and higher than the 2.6% of HIV positive PwE in the province (Nsanzimana et al., 2017;Van Steenkiste et al., 2019). It is noteworthy that no case of neurocysticercosis was reported, possibly due to low number of imaging studies and unavailability of diagnostic serological tests.
The relationship between epilepsy and perinatal complications, such as prolonged labor and birth asphyxia, has clearly been acknowledged (Osakwe et al., 2014). Five of 7 patients reporting perinatal complications, were male and accounted for more than 10% of male PwE. Total frequency of 5.5% was also higher than elsewhere reported (Osakwe et al., 2014).

4.3
Access to medical care, care seeking patterns, and gender The diagnosis gap was 62.5% with a 68.4% gap for female PwE and 51.0% for male PwE. We observed an overall treatment gap of 91.5% (117/128).
Financial reasons and access to care were not associated with gender differences, with a same percentage in men and women. Financial reasons were also evenly cited by both groups for either discontinuing treatment as for not seeking care. Professional status differed with more female reporting farmer as occupation, explained as an artifact given the age group distribution. Indeed, of all PwE above 20 y of age, about 15% of male and female reported an inactive or missing professional status. We did not document monthly income of PwE which may be a confounding factor. Yet, in the rural regions in Rwanda, most households are in the lowest economic class. As the three screened villages had a similar rural setting, we consider a bias due to economic status unlikely for diagnosis and treatment gap. We cannot exclude however that specific roles in the household, for example owning the financial budget, affect care seeking patterns. Geography itself has been reported to influence care seeking patterns, with patient reported proximity to the biomedical care center as a determining factor (Liu et al., 2019;Rutebemberwa et al., 2020). Our data did not allow an ascertainment of geographical distance as a determinant for gender differences.
For both diagnosis and treatment gap, a trend for gender differences was observed with female PwE citing beliefs and ignorance more frequently compared to male PwE. The latter mentioned more seizure and treatment related reasons. This observation was consistent between newly diagnosed and previously diagnosed patients. In our cohort, 22 PwE sought care with traditional healers or faith healers, of whom more than half did not seek biomedical care. Interestingly, this was more prevalent in male PwE, which has not been reported previously.
Proposed solutions to address the diagnosis and treatment gap, include involvement of CHWs to promote epilepsy care, social reconnection, and mutual support for PwE within communities (Mottiar & Lodge, 2018;Santos et al., 2019). CHWs are fully integrated and respected influencers in the village's core community activities. Second, social support groups play a vital role enabling community members to own their community healing practices, while maintaining existing social networks. Community's healing practices include initiatives, such as umusabane (social party for sharing life), umuganda (collective work to help vulnerable members), and umubugizi (mediation of reconciliation) (Otake, 2018). In our cohort, CHWs accompanied persons to the HC, and they stayed during the clinical investigation at the side of the positively screened person. Following the diagnosis and discussion of the proposed treatment, PwE returned with the CHWs in their communities. Under supervision of the CHW an early integration into the family and community was envisioned.

Limitations of this study
Incomplete screening of all villagers may induce an enrolment bias geared towards villagers staying at home, for example, children and women. Notwithstanding repeat visits, only 80% of villagers were screened, slightly lower than screening rates of 87% reported in a three-staged study design (Ngugi et al., 2013). Our distribution of screened villagers with 56% women reflected closely the census data with 54% female citizens and therefore our screening process is unlikely to have resulted in a bias towards gender or age overrepresentation. The gap of 20% does require an optimization of future screening processes. We cannot exclude a sample bias due to village selection, yet no significant differences in prevalence were observed.
Some of the observed differences from co-morbidities and epilepsy characteristics may be age-or gender-specific and therefore our observations must be interpreted with caution and considered as hypothesis generating only. We recommend future studies to include possible gender differences relative to age groups, aetiology, economic and sociocultural aspects.

CONCLUSION
We confirmed a high prevalence of epilepsy in rural regions in Rwanda

PEER REVIEW
The peer review history for this article is available at https://publons. com/publon/10.1002/brb3.2377

DATA AVAILABILITY STATEMENT
The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.